Jul 5, 2023 · 47m · mad
The Single Platform for Everyday AI - Fireside Chat with Florian Douetteau (Dataiku) & Matt Turck
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this fireside chat on The MAD Podcast, FirstMark Partner Matt Turck interviews Dataiku Co-Founder and CEO Florian Douetteau about scaling a $200M+ ARR enterprise AI platform, integrating Generative AI into existing data workflows, establishing responsible AI governance with the RAFT framework, and lessons learned on the founder-to-CEO journey.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Matt holds 19.7% of the talking time here. How this is scored →
Matt as informed peer 2.9 Guest teaching 5.1 Guest disagreement 0.4 Matt pushing back 0.4
speaking balance: gold is Matt, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 42:43 Florian corrects Matt on his technical title
In an otherwise exceptionally polite interview, Florian offers the softest correction of the transcript by noting he was never technically a data scientist, clarifying his background was in software engineering and product management.
Hardest push from Matt ▶ 28:37 Matt challenges GenAI adoption reality versus hypeMatt pushes past surface-level excitement to ask directly whether enterprise clients are implementing real production use cases or merely paying consultants for discovery mode.
Biggest teaching moment ▶ 6:56 Florian explains lazy compute and model inspectionFlorian delivers a thorough technical explanation of Dataiku's lazy pushdown compute architecture, automated model training, and the necessity of human inspection to distinguish causality from correlation.
Matt holds his own ▶ 19:20 Matt contextualizes enterprise AI timeline relative to ChatGPTMatt demonstrates sharp market awareness by contextualizing Dataiku's long history of deploying enterprise AI at scale against the recent post-ChatGPT public narrative.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Matt as informed peer | Guest teaching | Guest disagreement | Matt pushing back | Why |
|---|---|---|---|---|---|---|
| Dataiku's Centralized Platform Strategy and Founding Vision | 3 | 4 | 1 | 1 | Matt opens the episode with a background overview of Dataiku's funding and market position before asking about the centralized platform vision. Florian explains the founding logic of bridging domain experts and data scientists while avoiding complex tooling glue. Matt provides a succinct synthesis confirming how low-code and developer tools coexist. | |
| Comprehensive Product Tour: End-to-End Data to AI Lifecycle | 2 | 7 | 0 | 0 | Florian delivers an extensive product tour covering Dataiku 12, explaining pushdown compute, visual data preparation, AutoML selection, and MLOps. Matt briefly interjects to clarify how AutoML selects across the model spectrum. The dynamic is predominantly educational with Florian detailing technical architecture. | |
| Enterprise Use Cases and Vertical Industry Applications | 4 | 5 | 0 | 0 | Florian details industry use cases ranging from retail customer churn to shop floor sensor analytics replacing spreadsheets. Matt demonstrates solid domain perspective by highlighting how Dataiku deployed enterprise AI long before ChatGPT popularized the technology. Florian agrees and expands on enterprise AI frequency. | |
| Strategic Ecosystem Partnerships: Snowflake and Databricks | 2 | 5 | 0 | 0 | Matt prompts Florian on Dataiku's relationships with ecosystem partners Snowflake and Databricks. Florian outlines the historical evolution from Hadoop to cloud data warehouses, explaining Dataiku's position as an abstraction layer. | |
| Generative AI Capabilities and Real-World Enterprise Adoption | 3 | 6 | 1 | 2 | Matt asks how generative AI impacts Dataiku and presses on whether enterprise clients are seeing real production use cases or remain stuck in discovery mode. Florian explains structured versus unstructured data workflows and outlines concrete production examples in marketing, insurance contract extraction, and shop floor Q&A. | |
| AI Ethics, Governance, and the RAFT Framework | 4 | 5 | 0 | 0 | Matt references Dataiku's RAFT framework for AI ethics and governance. Florian details how large enterprises navigate regulatory uncertainty and reputational risk by evaluating accountability, fairness, transparency, and reliability ahead of formal compliance mandates. | |
| Future Trends in AI Interaction and Vertical Agent Builders | 2 | 5 | 0 | 0 | Florian describes future trends including multimodal UI interfaces combining code and natural language, alongside a new wave of domain hackers building specialized vertical agents without needing AI PhDs. | |
| Executive Leadership and the Founder-to-CEO Journey | 3 | 4 | 1 | 0 | Matt reflects on his 7-year investor relationship with Florian, praising his steady leadership style and asking how he transitioned from math and engineering to executive business management. Florian responds with personal humility and gently clarifies his exact background as a software engineer and product manager. |
Statements from this episode (10)
Dataiku reached over $200 million ARR and 1,200 employees
“That Taiku is 10 years old as a company. And in terms of numbers, we are at north of 1200 employees and we ended up last year north of two hundred million of revenue of annual recurring revenue.”
Git and code environments prevent business stakeholders from collaborating on data
“Git and coding environment, remove any possibility to have a meaningful discussion with people coming from the business, just because they won't get into code.”
Enterprise MLOps requires batch retraining and drift management, not just APIs
“A misconception is that moving a model in projection is about turning a model into an API, which is of course, one step, but also in many instances, it's about robustifying models that can be applied in a batch fashion, robustifying the fact of automating the …”
Simple ML models on real data easily outperform traditional business rules
“In many businesses applying models that are actually not so complicated, but like, still being, you know, with real life data and with a large amount of data, you outperform your previous business rule by a significant margin.”
Most companies will run hundreds of AI automation systems within ten years
“I think that in the next five to 10 years, it will be actually Universally mainstream, as in most companies, a very vast majority of companies will have hundreds, if not thousands of automated systems that automate ordering and some form of cash management and…”
Generative AI fragmentation mirrors the data warehousing market ten years ago
“There is some parallel between the current situation of generative AI and the situation of data warehousing back 10 years ago, as in a multiplicity of technologies and multiplicity of models, question marks in terms of how you integrate them together.”
Companies will adopt AI governance and self-assessments ahead of formal regulation
“I think that many companies will have to adopt this kind of systems a little bit ahead of time in order to manage some of the internal expectation in terms of control of AI or expectation on their customers, but also in order to be more ready when the complian…”
Future enterprise software will seamlessly combine visual, text, and code interfaces
“I think that there would be, there is a future of, let's say, Enterprise software where you've got system of exist mix of visual and text and potentially coding representation of things where you can navigate between code and clicking and talking to your compu…”
Generative AI empowers non-PhD software builders to create specialized vertical agents
“I think that what I see as very interesting with generative AI is that it's like a new Lego block enabling a new generation of hackers for many vertical applications. And there are like many, many, many, many vertical applications that can be built using and b…”
Technical founders should not attempt to become sales or communications experts
“When you come from a technical technical background, you won't become like a service person or a communication person. They want, and maybe it's actually not an objective to become. So you have to stay a little bit what you are, because ultimately you have to …”